FIX: AutoPeftModels never reduce embedding size#2427
Merged
BenjaminBossan merged 1 commit intohuggingface:mainfrom Mar 14, 2025
Merged
Conversation
Resolves huggingface#2415 There was a bug in AutoPeftModels where the embedding was always resized to the vocab size of the tokenizer when the tokenizer was found. This makes sense if the vocabulary was extended, but some models like Qwen already start out with "spare" embeddings, i.e. the embedding size is larger than the vocab size. This could result in the embedding being shrunk, which in turn resulted in an error when loading the weights.
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
Member
Author
|
MacOS errors are unrelated and seem to stem from this accelerate PR: huggingface/accelerate#3420. |
githubnemo
approved these changes
Mar 14, 2025
Guy-Bilitski
pushed a commit
to Guy-Bilitski/peft
that referenced
this pull request
May 13, 2025
Resolves huggingface#2415 There was a bug in AutoPeftModels where the embedding was always resized to the vocab size of the tokenizer when the tokenizer was found. This makes sense if the vocabulary was extended, but some models like Qwen already start out with "spare" embeddings, i.e. the embedding size is larger than the vocab size. This could result in the embedding being shrunk, which in turn resulted in an error when loading the weights.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Resolves #2415
There was a bug in
AutoPeftModels where the embedding was always resized to the vocab size of the tokenizer when the tokenizer was found. This makes sense if the vocabulary was extended, but some models like Qwen already start out with "spare" embeddings, i.e. the embedding size is larger than the vocab size. This could result in the embedding being shrunk, which in turn resulted in an error when loading the weights.